76189 MARCH 2013 • Number 111 Global Connectivity and Export Performance Jean-François Arvis and Ben Shepherd The World Bank has developed a novel method for measuring countries’ connectivity in global networks and has applied it to the global air transport network. “Connectivity� in this context is defined as a country’s relative position in that network in terms of the total “push� and “pull� it exerts on air traffic, taking account of all possible links with other countries. Well-connected countries that are strongly connected to other well-connected countries are considered “hubs� in this definition. Less well-connected countries are “spokes.� The Air Connectivity Index (ACI) shows that connectivity is highly concentrated in North America and Europe (“hubs�); most developing countries are relatively poorly connected (“spokes�). Developing countries looking to increase their participation in global value chains need to improve their connectivity as part of their overall competitiveness strategy, including the progressive liberalization of their air transport sectors. What Is Connectivity? global and regional networks in areas such as transport, fi- nance, and trade is a key determinant of its degree of integra- One of the most salient features of the international economy tion into the international economy. “Connectivity� in this over recent years has been the rise of networked production note refers to a country’s ability to effectively connect to oth- through global and regional value chains. In many parts of the ers within a particular network. The concept of connectivity is world, intermediate inputs are manufactured at a location thus key in the context of global and regional value chains. Put different from the place of final assembly, which means that another way, looking at traffic patterns, is a country a “hub� or goods cross borders frequently before finally being shipped to a “spoke,� and does it trade with “hubs� or “spokes�? Countries a remote consumer. Electronics is the sector for which global that are becoming more competitive in this context are also value chains are most well-known, including products such as working hard to improve their connectivity. In some parts of the iPhone. But global value chains also exist in other sectors, the world, regional cooperation—such as the Asia-Pacific Eco- such as textiles and clothing, and even processed agriculture. nomic Cooperation’s Supply Chain Connectivity Initiative—is As a result, a wide range of countries at various income levels used as a specific platform to improve connectivity. are all involved in globalized or regionalized production plat- The most important innovation in viewing economic ac- forms. tivity through the twin lenses of global and regional produc- For countries still on the edge of global value chains, the tion networks and connectivity is that production and con- key question is how to effectively integrate them to reap the sumption are no longer viewed as a point-to-point economic economic benefits that come with increased trade and foreign interaction. Instead, the unit of analysis becomes the network, investment. Because goods move so frequently and so far with- which is inherently a nonlinear framework. An important as- in the value chain model, a country’s ability to connect to pect of the competitiveness agenda therefore becomes im- 1 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise proving a country’s position in global networks not just of ticular network, Arvis and Shepherd (2011) defined connec- trade and production, but also in networks that support those tivity as the total “push� or “pull� exerted by a country on the processes, such as air and maritime transport and finance, global air transport network, taking into account all possible among others. connections—even those where there are no actual flights Connectivity is a difficult concept to formalize and assess available. The World Bank summarizes connectivity in a sin- in a useful way across countries. One example is provided by gle number, the Air Connectivity Index (ACI). Countries Arvis and Shepherd (2011), who analyzed connectivity in the have a higher ACI score and are better connected to the net- global air transport industry. This note builds on their analy- work if they have strong connections to other countries that sis and provides a robust methodology for measuring connec- are themselves well connected, and if the dispersion of their tivity, firmly grounded in economic theory. It also shows that connectivity scores is low. Thus, countries with relatively connectivity is a key determinant of important economic out- strong connections to a wide range of countries that are them- come variables, such as the percentage of parts and compo- selves well connected tend to have higher ACI scores. Those nents in total exports, which is one indicator of the extent of countries with relatively weak connections to countries that a country’s integration into global supply chains. are themselves poorly connected record lower ACI scores. The ACI itself is normalized as a percentage of the theoretically Measuring Connectivity maximum global push or pull: zero indicates that a country has no connections whatsoever, and 100 percent indicates A number of different disciplines have considered the con- that it is strongly and directly connected to all other countries cept of connectivity, but few have applied it to international in the network. trade. Definitions change according to the perspective taken The theoretical background for the ACI is a standard and the end use envisaged. In applied mathematics (“network gravity model, taken from the trade and regional science lit- theory�), for example, the purpose of measuring connectivity eratures. The formulation used is very flexible and is consis- is primarily to describe the overall hierarchy of a network, tent with a variety of microeconomic underpinnings. As such as a subset of the Internet, and identify the most central such, the ACI represents the first attempt to measure connec- points (“nodes�). The core techniques are based on linear alge- tivity in a global network using a theoretically grounded ap- bra, such as the widely used eigenvector centrality (a modifica- proach. The same approach could conceivably be applied in tion of which is used to run Google’s Page Rank algorithm), other areas such as maritime transport, trade, and finance. which is primarily a way of summarizing an important charac- teristic of real-world networks. However, network theory lit- Connectivity in the Global Air Transport erature does not have any direct economic policy application. Network By contrast, the policy literature—such as the United Na- Application of the ACI model to data on global air transport tions Conference on Trade and Development’s (UNCTAD) flows covering 211 countries and territories for the year 2007 Liner Shipping Connectivity Index—targets policy makers produces stylized facts that are highly consistent with expec- and analysts for its main audience. The UNCTAD approach tations. As figure 1 shows, there are a small number of rela- consists of taking a weighted average of a certain number of tively well-connected countries, but most countries remain countrywide indicators of economic and shipping activity connectivity, which are then summarized in a single index. This heuristic approach has proved very useful in applied Figure 1. Few Countries Have High ACI Scores (Hubs), Most Are work, but is not rooted in theory and is not explicitly ground- Poorly Connected (Spokes) ed in network analysis. The challenge for policy-relevant work on connectivity in .2 the context of international economics is therefore to develop a concept that provides a network understanding in which .15 the interaction between two countries depends on all other density interactions in the network. Unlike linear models from stan- .1 dard “network theory,� the concept should also include the existence of links between countries, as well as the mecha- nism that explains the intensity of transport or trade flows .05 along those links. Air Connectivity Index 0 0 5 10 15 20 Moving from the more general concept of connectivity as a Air Connectivity Index country’s ability to effectively connect to others within a par- Source: Authors’ calculations. 2 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise relatively poorly connected. This finding matches the hub- measured by the ACI and this proxy for participation in and-spoke nature of the global air transport network, a feature global value chains. that has been widely emphasized in industry literature. In line with this finding, recent work by Arvis et al. The ACI shows that the United States is the most con- (2013) shows that air and maritime connectivity are impor- nected country, with a score of around 23 percent. There is tant determinants of the level of trade costs between coun- then a substantial gap between the United States and the next tries. Global and regional value chains can only flourish in group of countries, namely Canada and the main European environments with low trade transaction costs. New data on hubs. Third-ranked Germany, for example, has an ACI score trade costs in the developing world for 1995–2010 show that of just over 12 percent. The developing country with the connectivity and logistics performance are at least as impor- highest score is Algeria, with just under 7 percent, because of tant as geographical distance in determining the level of trade its strong connections with European hubs. The main hubs in costs between countries. Asia and the Middle East do not score as highly as might be expected, because they are more distant from the major mar- Conclusion and Policy Implications kets of the United States and Europe. Trade policy, particularly “trade and transport facilitation,� Connectivity, Policy, and Trade Outcomes the set of policies designed to reduce trade transaction costs, increasingly needs to focus on networks of trade and produc- How do countries become better connected in the global air tion as the analytical unit of primary interest. Such an ap- transport network? Policy is of course a major factor. Build- proach naturally leads to increased attention on connectivity ing infrastructure and putting in place well-functioning as a determinant of competitiveness in a networked world. markets for backbone services are crucial. But so too is the Improved connectivity can substantially reduce the transac- degree of liberalization of air transport markets. Figure 2 tion costs associated with exporting and importing, and shows that there is a strong positive correlation between lib- thereby improve a country’s ability to take full advantage of eralization as measured by the World Trade Organization’s the benefits offered by global and regional value chains. (WTO) Air Liberalization Index (ALI) and connectivity, as In the context of the air transport network, the World measured by the ACI. Bank’s ACI provides a robust and theoretically grounded In line with the view advanced earlier, that connectivity methodology for measuring connectivity. Its structure sug- is a key determinant of competitiveness in the context of gests that policy makers need to be concerned with two pri- global and regional value chains, evidence also shows a strong mary factors when it comes to connectivity: (i) building connection between the ACI and an important trade indica- stronger links with global and regional hubs and (ii) increas- tor of value chain participation (figure 3). Although there is ing the number and quality of connections with a wide range no direct evidence on the extent to which different coun- of countries to improve their place in the global network. Lib- tries participate in production based on global value chains, eralization of air services markets—which takes place primar- a common proxy measure is the percentage of machinery ily on a bilateral basis—has an important role to play in im- parts and components in total exports. As expected, there is a strong positive correlation between air connectivity as Figure 3. ACI Positively Correlated with Trade in Global Value Chains Figure 2. The Positive Correlation between Air Transport 25 Liberalization and Connectivity .25 20 .2 15 percent .15 10 ACI .1 5 .05 0 0 .05 .10 .15 .20 .25 0 ACI 0 10 20 30 40 parts and components in total exports (%) mean ALI �tted values ACI �tted values Source: Authors’ calculations. Note: Trade in global value chains is proxied by the percentage of machinery parts Source: Authors’ calculations. and components in total exports. 3 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise proving connectivity, as does the improvement of References infrastructure and capacity building in the area of aviation Arvis, J.-F., Y. Duval, B. Shepherd, and C. Utoktham. 2013. “Trade and related services. Moving forward on connectivity in de- Costs in the Developing World: 1995–2010.� Policy Research veloping countries requires policy action on a number of Working Paper No. 6309, World Bank, Washington, DC. fronts, but it has the potential to provide a powerful boost for Arvis, J.-F., and B. Shepherd. 2011. “The Air Connectivity Index: efforts to improve export competitiveness. Measuring Integration in the Global Air Transport Network.� Policy Research Working Paper No. 5722, World Bank, Wash- About the Authors ington, DC. Jean-François Arvis is a Senior Transport Economist and Ben Shepherd is a consultant; both are with the International Trade Department (PRMTR) at the World Bank. The Economic Premise note series is intended to summarize good practices and key policy findings on topics related to economic policy. They are produced by the Poverty Reduction and Economic Management (PREM) Network Vice-Presidency of the World Bank. The views expressed here are those of the authors and do not necessarily reflect those of the World Bank. The notes are available at: www.worldbank.org/economicpremise. 4 POVERTY REDUCTION AND ECONOMIC MANAGEMENT (PREM) NETWORK    www.worldbank.org/economicpremise